Floating Point Library for CC2650

Compiler will just use a library function that uses integer math to emulate floating point math.

That is my concern.
A simple filter with 8 multiplications and additions takes about 40 microseconds.
That is about 5 microseconds for 1 multiplication and 1 addition.
This is not good enough. So I was asking about a specialized fp library.
Just for comparison, Dialog's new chip is 2 to 3 times quicker.

Hi Aharon,

As YiKai suggested using actual floats is discouraged on this device as they will be emulated. I recommend using fixed point arithmetic, which should be able to offer the accuracy you need without the performance degradation.

Hi Sean,

Where and how can I enable/install the fixed point math library for CC13xx/CC26xx device? I am using CCS v6.1 for development. 

../ming

Fixed point math is used by default and you don't have to install anything.

Hi YiKai!

Could you please elaborate on that?

Thanks!

Can you specify what you want me to elaborate?

About the fixed point library for the CC2650. So far the only references to working with fixed point values that I have found are for the MSPXX series using the IQmath. You said that the CC2650 automatically uses fixed point, but I can't find any references explaning how to use it.

Thanks in advance!